Ok, the point is not the mind reading, only that we can detect something going on in the brain.
We need subjects/participants to TELL us about what they are doing (or at least assume they are following the tasks) in order to get even basic interpretations of neural correlate data off of the ground. What we can't do is make the leap from the assumption that brain imaging data is correlated with particular cognitive processes to the conclusion all cognitive processes or consciousness or the mind is physical (apart from
a priori assumptions).
It can be as simple as being able to detect when someone is wake vs asleep. Something is different and that difference is detectible therefore physical.
By that logic, I've seen people "detect" the presence of ghosts and other paranormal activity.
However, do you have an answer to the question in the OP?
What non-physical do you accept?
I don't think it makes much sense. In physics, we deal with things all of the time that we don't really regard as being "physical" in the same sense the term is usually used colloquially. A lot of "fundamental" physical processes and constituents are in fact book-keeping devices for extracting data from experiments in HEP physics or in other similar arenas. Elsewhere, the problem is circumvented by using terms like "information" which is claimed to be more fundamental and treated as such because it is abstract to begin with and we don't have to worry about metaphysics.
We still make claims, however, about the very physical nature of things like probability currents and probability conservation, which is not physical in any meaningful sense but absolutely vital to our understanding of fundamental physics and its foundations.
On the other hand, absolutely basic components of physical experiences are excluded
a priori from physics because we need to do experiments, Thus we need to pretend that we can set initial conditions for physical systems (with or without interaction or environment) in an idealized manner.
As a result, something as basic as "time" enters into fundamental physics as a parameter (even when we have to rename it as e.g., proper time because we need a temporal dimension to physical space).
Most of what is called "physical" in HEP, quantum many-body physics, quantum field theories, etc., consists of virtual processes and particles that are not physical in the sense the term is usually meant.
Additionally, the governing laws themselves (as well as governing principles such as the action) are postulated to have real physical consequences at cosmic and microcosmic scales but to have no physical existence.
